Degradation of Serotonin N-Acetyltransferase, a Circadian Regulator, by the N-end Rule Pathway*

Serotonin N-acetyltransferase (AANAT) converts serotonin to N-acetylserotonin (NAS), a distinct biological regulator and the immediate precursor of melatonin, a circulating hormone that influences circadian processes, including sleep. N-terminal sequences of AANAT enzymes vary among vertebrates. Mechanisms that regulate the levels of AANAT are incompletely understood. Previous findings were consistent with the possibility that AANAT may be controlled through its degradation by the N-end rule pathway. By expressing the rat and human AANATs and their mutants not only in mammalian cells but also in the yeast Saccharomyces cerevisiae, and by taking advantage of yeast genetics, we show here that two “complementary” forms of rat AANAT are targeted for degradation by two “complementary” branches of the N-end rule pathway. Specifically, the Nα-terminally acetylated (Nt-acetylated) Ac-AANAT is destroyed through the recognition of its Nt-acetylated N-terminal Met residue by the Ac/N-end rule pathway, whereas the non-Nt-acetylated AANAT is targeted by the Arg/N-end rule pathway, which recognizes the unacetylated N-terminal Met-Leu sequence of rat AANAT. We also show, by constructing lysine-to-arginine mutants of rat AANAT, that its degradation is mediated by polyubiquitylation of its Lys residue(s). Human AANAT, whose N-terminal sequence differs from that of rodent AANATs, is longer-lived than its rat counterpart and appears to be refractory to degradation by the N-end rule pathway. Together, these and related results indicate both a major involvement of the N-end rule pathway in the control of rodent AANATs and substantial differences in the regulation of rodent and human AANATs that stem from differences in their N-terminal sequences.

. Regulatory functions of NAS include its activity as an agonist of TrkB, the receptor for the brain-derived neurotrophic factor (12)(13)(14). NAS is the immediate precursor of melatonin, a circulating hormone that regulates sleep and other circadian processes in vertebrates. Melatonin is also present in invertebrates, including insects, as well as in animals that lack recognizable neurons (1,(15)(16)(17)(18)(19)(20)(21). The functions of melatonin in mammals include the modulation of circadian rhythms in response to light-dark cycles. Melatonin also contributes to the control of seasonal physiology (20). These mechanisms are a part of a broader range of processes that involve distinct biological oscillators in all organisms (22)(23)(24). In the course of daily light-dark cycles, the activity of AANAT and the level of the ϳ23-kDa AANAT protein in the brain's pineal gland are high during nocturnal periods and rapidly decrease following exposure to light (1,20,25).
In mammals, changes of AANAT levels in the brain's pineal gland are controlled by the circadian oscillator in the suprachiasmatic nucleus of the hypothalamus. At night, axons of rodent superior cervical ganglion neurons release norepinephrine in the pineal gland, in response to circadian signals from the suprachiasmatic nucleus, thereby activating adrenergic receptors to increase intracellular Ca 2ϩ and cAMP. The resulting phosphorylation of CREB, the cAMP-response element-binding protein, up-regulates the Aanat transcriptional promoter, which contains cAMP-response element sequence elements (1,4,(25)(26)(27). Increases in cAMP also stimulate the site-specific phosphorylation of AANAT by PKA, thereby causing the binding of AANAT to 14-3-3 proteins. These transitions augment the enzymatic activity of AANAT and inhibit its degradation (1,6,7,28). Although AANAT is predominantly expressed in the pineal gland and the retina, this enzyme is also present in cells of the gastrointestinal tract and apparently at other sites as well, including the hippocampus, the olfactory bulb, the cerebellum, and the spinal cord (12,29,30).
Melatonin is produced from NAS by hydroxyindole-Omethyltransferase. The levels of NAS, synthesized by AANAT, can constrain the rate of melatonin synthesis from NAS during light parts of daily cycles (when NAS levels are low) but appear not to be the rate-limiting step during dark parts of these cycles, when both AANAT and NAS levels are high (21). In rodents such as rats, the activity of AANAT starts to increase 3-4 h after the onset of darkness and begins to decrease before the onset of light in the morning (21,31). Hence the importance of regulating the expression and the enzymatic activity of AANAT, because these parameters determine the levels of both NAS and melatonin.
The concentration of AANAT, a largely cytosolic enzyme (32), is controlled through both transcriptional and post-translational mechanisms (1,11,31,(33)(34)(35). In rodents, the onset of darkness leads to an increase in melatonin (requiring a preceding increase of NAS) after a lag period, whereas in sheep and primates, increases in melatonin occur rapidly upon the onset of darkness. The levels of Aanat mRNA in rodents can vary by Ͼ100-fold during light-dark cycles (10), whereas in nonrodent mammals such as, for example, sheep, these differences can be as low as ϳ2-fold, suggesting a major involvement of translational and/or post-translational AANAT regulation in latter cases (11,36). In addition to variability in regulation at the level of AANAT-encoding mRNAs, AANAT is also unusual in the extent of variability of its N-terminal sequences during vertebrate evolution (Fig. 1C). In 2010, one of our laboratories showed that N-terminal residues of rat AANAT play a role in the control of its degradation, suggesting the involvement of the N-end rule pathway (37).
The N-end rule pathway is a set of intracellular proteolytic systems whose unifying feature is the ability to recognize and polyubiquitylate proteins containing N-terminal degradation signals (degrons) called N-degrons, thereby causing degradation of these proteins by the proteasome (Fig. 1, A and B) (38 -46). Ubiquitin (Ub) ligases of the N-end rule pathway, called N-recognins, can recognize not only N-degrons but also specific internal degradation signals (47)(48)(49)(50). The main determinant of an N-degron is either an unmodified or chemically modified destabilizing N-terminal residue of a protein. Another determinant of an N-degron is a protein's internal Lys residue(s). It functions as the site of polyubiquitylation and tends to be located in a conformationally disordered region (41,51,52). Bacteria also contain the N-end rule pathway, but Ub-independent versions of it (53)(54)(55)(56)(57)(58).
Regulated degradation of proteins and their natural fragments by the N-end rule pathway has been shown to mediate a strikingly broad range of biological functions, including the sensing of heme, nitric oxide (NO), oxygen, and short peptides; the control of the input stoichiometries of subunits in oligomeric protein complexes; the elimination of misfolded or otherwise abnormal proteins; the degradation of specific proteins after their retrotranslocation to the cytosol from membrane-enclosed compartments; the regulation of apoptosis and repression of neurodegeneration; the regulation of DNA repair, transcription, replication, and chromosome cohesion/segregation; the regulation of G proteins and cytoskeletal proteins, such as actin and myosin; the regulation of autophagy, peptide import, meiosis, immunity, fat metabolism, cell migration, cardiovascular development, spermatogenesis, and neurogenesis; the functioning of adult organs, including the brain, muscle, testis, and pancreas; and the regulation of leaf and shoot development, leaf senescence, and many other processes in plants (Fig. 1, A and B) (see Refs. 41-46 and references therein).
In eukaryotes, the N-end rule pathway consists of two branches. One branch, called the Ac/N-end rule pathway, targets proteins for degradation through their N ␣ -terminally acetylated (Nt-acetylated) residues ( Fig. 1B) (39, 40, 46, 59 -62). Degradation signals and E3 Ub ligases of the Ac/N-end rule pathway are called Ac/N-degrons and Ac/N-recognins, respectively. Nt-acetylation of cellular proteins is apparently irreversible, in contrast to acetylation-deacetylation of proteins' internal Lys residues. About 90% of human proteins are cotranslationally Nt-acetylated by ribosome-associated Ntacetylases (63). Ac/N-degrons are present in many, possibly most, Nt-acetylated proteins (Fig. 1B). Natural Ac/N-degrons are regulated through their reversible shielding in cognate protein complexes (59).
In S. cerevisiae, the Ac/N-end rule pathway is mediated by (at least) the cytosolic/nuclear E3 Ub ligase Not4 and by Doa10, an endoplasmic reticulum membrane-embedded E3 (59). In mammalian cells, this pathway is mediated by (at least) the Teb4 E3, which is sequelogous (similar in sequence (71)) to yeast Doa10 (60). Human RGS2, a regulator of specific G proteins, is a shortlived substrate of the Ac/N-end rule pathway both in mammalian cells and in the heterologous setting of S. cerevisiae (60,72,73). In addition, the naturally occurring (blood pressure-elevating) human RGS2 Q2L mutant (in which Gln at position 2 is replaced by Leu) is targeted for degradation by both the Ac/Nend rule pathway and the Arg/N-end rule pathway (60). The Ac/N-end rule pathway recognizes the Nt-acetylated Ac-RGS2 Q2L (specifically its N-terminal Ac-Met residue), whereas the non-Nt-acetylated RGS2 Q2L is targeted (through its N-terminal Met-Leu sequence) by the Arg/N-end rule pathway (60).
In the present work, we analyzed the proteasome-mediated degradation of rat AANAT (37), whose N-terminal sequence Met-Leu (Fig. 1C) is identical to that of the otherwise unrelated human RGS2 Q2L protein (60). We also characterized human AANAT, which is highly sequelogous (71) (84% identical) to rat AANAT but bears a different N-terminal sequence (Fig. 1C). Our analyses employed the previously helpful approach of dissecting degradation of a mammalian protein of interest not only in a homologous (mammalian) setting but also in S. cerevisiae (60,66), thereby making possible the use of yeast genetics.
We show here that two alternative versions of rat AANAT, its Nt-acetylated and non-Nt-acetylated forms, are targeted for degradation by the Ac/N-end rule pathway and the Arg/N-end rule pathway, respectively. In contrast, human AANAT, whose N-terminal sequence differs from that of rodent AANATs, is significantly longer-lived than its rat counterpart and appears to be largely refractory to degradation by the N-end rule pathway. Together, these and related results indicate both a major involvement of the N-end rule pathway in the control of rodent AANATs and substantial differences in the regulation of rodent and human AANATs that stem from differences in their N-terminal sequences.

Wild-type Rat and Human AANATs and Their Mutants-
The start (AUG) codon-encoded N-terminal Met residue of nascent proteins is cotranslationally cleaved off by ribosomeassociated Met-aminopeptidases (MetAPs) if a residue at position 2, to be made N-terminal by the cleavage, is not larger than Val (41,74). Thus, for example, the Met residue of the Nterminal Met-Leu-Ser sequence of wild-type rat AANAT (denoted as MLS rAANAT) (Fig. 1C) is retained in mature MLS rAANAT, inasmuch as Leu is larger than Val. Given the previously demonstrated usefulness of employing genetic tractability of S. cerevisiae for understanding the targeting of mammalian N-end rule substrates (60,66), we carried out the present study by expressing AANAT test proteins not only in human HEK293T cells but also in S. cerevisiae. Cited below are the examined wild-type and mutant AANAT proteins.
When present, parentheses around a superscript's N-terminal Met residue in the notations of AANAT test proteins denote the fact that this Met is cotranslationally cleaved off by MetAPs.
(ii) (M)SMLS rAANAT 3f , in which the sequence (Met)-Ser was placed before the wild-type Met-Leu-Ser sequence of rat MLS rAANAT 3f , yielding the N-terminal sequence (Met)-Ser-Met-Leu-Ser. Its N-terminal Met residue would be cotranslationally cleaved off by MetAPs ( Fig. 2A). The (Met)-Ser sequence was added to wild-type MLS rAANAT 3f in order to mimic the (Met)-Ser N-terminal sequence of primate (including human) AANATs (Fig. 1C).
(iii) (M)SI rAANAT 3f , in which Leu at position 2 of the wildtype rat MLS rAANAT 3f (Fig. 1C)  (iv) (M)PLS rAANAT 3f , in which the Pro residue was inserted between N-terminal Met and second-position Leu of wild-type rat MLS rAANAT 3f . This alteration aimed to address the relevance of Nt-acetylation to the degradation of wild-type rat MLS rAANAT 3f . Specifically, the N-terminal sequence Met-Leu-Ser is a priori likely to be cotranslationally Nt-acetylated in vivo at its (retained) N-terminal Met (59,62,63). In contrast, the N-terminal Met residue of the mutant N-terminal sequence (Met)-Pro-Leu-Ser of (M)PLS rAANAT 3f would be cotransla-tionally cleaved off by MetAPs. The resulting N-terminal Pro is not Nt-acetylated, at least in S. cerevisiae, and is usually not Nt-acetylated in mammalian cells as well (59,63).
(v) MLS rAANAT 3ha , the wild-type rat MLS rAANAT ( Fig. 1C) C-terminally tagged with a triple-HA epitope. It was used as a control for the MLS rAANAT 3ha K8R mutant described in item viii. (vi) (M)ST hAANAT 3f , the wild-type human (M)ST hAANAT ( Fig. 1C) C-terminally tagged with a triple-FLAG epitope.
(vii) (M)PT hAANAT 3f , in which the second-position Ser of the wild-type human (M)ST hAANAT 3f (Fig. 1C) was replaced by the Pro residue, yielding (M)PT hAANAT 3f . The N-terminal Met residue of either the wild-type human (M)ST hAANAT 3f or the mutant (M)PT hAANAT 3f would be cotranslationally removed by MetAPs (41,74). However, in contrast to the resulting N-terminal Ser of wild-type ST hAANAT 3f , which would be expected to be cotranslationally Nt-acetylated, the N-terminal Pro of the mutant PT hAANAT 3f is not Nt-acetylated, at least in S. cerevisiae, and is usually not Nt-acetylated in mammalian cells as well (59,63).
(viii) MLS rAANAT 3ha K8R , the otherwise wild-type triple-HAtagged rat MLS rAANAT in which the Lys-8 residue was replaced by Arg.
(ix) MLS rAANAT 3ha Kzero , the otherwise wild-type triple-HAtagged rat MLS rAANAT in which each of its four Lys residues were replaced by Arg (Fig. 6B).
(x) (M)ST hAANAT 3ha Kzero , the otherwise wild-type triple-HAtagged human (M)ST hAANAT in which both of its Lys residues were replaced by Arg (Fig. 7, A and D).
Because the N-terminal Met residue in the Met-Leu-Ser sequence of the wild-type rat MLS rAANAT 3f would be likely to be Nt-acetylated in vivo (59,62,63), the rapid degradation of MLS rAANAT 3f (t1 ⁄ 2 Ͻ 5 min) (Fig. 2, B and D) suggested the involvement of the Ac/N-end rule pathway (Fig. 1A). To address this possibility, we constructed the non-Nt-acetylatable (M)PLS rAANAT 3f mutant (see item iv above). (This mutant would also not be expected to be targetable by the Arg/N-end rule pathway, because the latter does not recognize N-terminal Pro (40,41).) Remarkably, (M)PLS rAANAT 3f was long-lived in S. cerevisiae, in comparison with the short-lived wild-type MLS rAANAT 3f (t1 ⁄ 2 Ϸ 1 h versus t1 ⁄ 2 Ͻ 5 min, respectively), strongly suggesting (but not proving) a role for the Ac/N-end rule pathway in the degradation of the wild-type rat MLS rAANAT 3f (Fig. 2, B (lanes 15-18) and D; also see below).
Crucially, the relative stabilization of MLS rAANAT 3f was dramatically higher in the double mutant naa30⌬ ubr1⌬ than in either one of the single mutants ( Fig. 2, C (lanes 18 -21) and E). These findings should be considered together with the conceptually independent result that the non-Nt-acetylatable (and Amino acid residues are denoted by single-letter codes. A, the mammalian Arg/N-end rule pathway. It targets proteins for degradation through their specific unacetylated N-terminal residues. A yellow oval denotes the rest of a protein substrate. Primary, secondary, and tertiary, mechanistically distinct classes of destabilizing N-terminal residues. Ntan1 and Ntaq1, N-terminal amidases (Nt-amidases) that convert the tertiary destabilizing N-terminal residues Asn and Gln to Asp and Glu, respectively. The Ate1 arginyltransferase (R-transferase) conjugates Arg, a primary destabilizing residue, to N-terminal Asp, Glu, and (oxidized) Cys. Type 1 and type 2, two sets of primary destabilizing N-terminal residues, basic (Arg, Lys, and His) and bulky hydrophobic (Leu, Phe, Trp, Tyr, Ile, and Met, if the latter is followed by a bulky hydrophobic residue (⌽)), respectively. These sets of N-terminal residues are recognized by two distinct substrate-binding sites of N-recognins, the pathway's E3 ubiquitin ligases, whose (possibly incomplete) list includes Ubr1, Ubr2, Ubr4, and Ubr5. B, mammalian Ac/N-end rule pathway. It targets proteins through their N ␣ -terminally acetylated (Nt-acetylated) residues. Red arrow on the left, cotranslational removal of the N-terminal Met residue by Met-aminopeptidases (MetAPs). N-terminal Met is retained if a residue at position 2 is larger than Val. C, the first 10 amino acid residues of animal AANAT enzymes, in vertebrates and an invertebrate, such as Drosophila melanogaster. See the Introduction for descriptions of AANATs.
also not targetable by the Arg/N-end rule pathway) (M)PLS rAANAT 3f mutant (see the preceding paragraph) was much longer-lived in wild-type S. cerevisiae than its wild-type MLS rAANAT 3f counterpart (Fig. 2, B (lanes 15-18 versus lanes [3][4][5][6] and D). Together, these sets of findings indicated that the wild-type rat MLS rAANAT 3f (its two alternative forms) was targeted for degradation by both the Ac/N-end rule pathway and the Arg/N-end rule pathway. Specifically, the Nt-acetylated Ac-MLS rAANAT 3f is destroyed through the recognition of its Nt-acetylated N-terminal Met residue by the Ac/N-end rule pathway (Fig. 1B), whereas the non-Nt-acetylated MLS rAANAT 3f is targeted by the "complementary" Arg/N-end rule pathway, which recognizes the unacetylated N-terminal Met-Leu sequence of MLS rAANAT 3f as the Met-⌽ motif (N-terminal Met followed by a bulky hydrophobic residue) (Fig. 1A).
This conclusion was in agreement with the finding that the rapid degradation of MLS rAANAT 3f in wild-type S. cerevisiae FIGURE 2. The wild-type rat MLS rAANAT as a substrate of both branches of the N-end rule pathway. A, the wild-type rat MLS rAANAT and its mutants that were constructed and analyzed in the present study. Amino acid residues are denoted by single-letter codes. The Leu residue, at position 2 in wild-type MLS rAANAT, is highlighted in red to make it easier to follow sequence alterations in specific mutants. B, immunoblotting with antibody to actin was used as a loading control. Lane 1, fluorescently labeled molecular mass markers (LI-COR); their masses, in kDa, are indicated on the left. Lane 2, wild-type S. cerevisiae were transformed with vector (V) alone (control). CHX-chases were performed at 30°C for the indicated times in wild-type S. cerevisiae with the wild-type rat MLS rAANAT 3f (lanes 3-6) and with its mutants (M)SMLS rAANAT 3f (lanes 7-10), (M)SI rAANAT 3f (lanes [11][12][13][14], and (M)PLS rAANAT 3f (lanes [15][16][17][18]. The bands of AANAT 3f and actin are indicated on the right. C, immunoblotting with antibody to tubulin was used as a loading control. Lane 1, 37 and 50 kDa molecular mass markers (see lane 1 in B). Shown are CHX-chases with the wild-type rat MLS rAANAT 3f in wild-type S. cerevisiae (lanes 2-5) and its mutants naa30⌬ (lanes 6 -9), naa10⌬ (lanes 10 -13), ubr1⌬ (lanes 14 -17), naa30⌬ ubr1⌬ (lanes 18 -21), and naa10⌬ ubr1⌬ (lanes [22][23][24][25]. The bands of MLS rAANAT 3f and tubulin are indicated on the left. D, quantification of data in B. E, quantification of data in C. All quantified CHX-chase assays were carried out at least three times and yielded results within 10% of the data shown.

Degradation of AANAT by the N-end Rule Pathway
(t1 ⁄ 2 Ͻ 5 min) remained unchanged in naa10⌬ cells, which lacked the non-cognate NatA Nt-acetylase (which does not Ntacetylate N-terminal Met) (59,76). Furthermore and also in agreement with the above conclusion, the degradation of MLS rAANAT 3f in double-mutant naa10⌬ ubr1⌬ cells proceeded at a rate similar to that in the single ubr1⌬ mutant (Fig. 2, C (lanes 22-25) and E). Thus, a double-mutant background that nearly completely stabilizes the wild-type rat MLS rAANAT 3f must be naa30⌬ ubr1⌬ in that it should also lack the activity of the cognate NatC Nt-acetylase ( Fig. 2, C (lanes 18 -21) and E). In agreement with this conclusion, the ablation of NAA10, encoding the non-cognate NatA Nt-acetylase, did not have a synergistic effect on the rate of MLS rAANAT 3f degradation in double-mutant naa10⌬ ubr1⌬ cells (Fig. 2

, C (lanes 22-25) and E).
Previous work has identified the endoplasmic reticulum membrane-embedded E3 Ub ligase Doa10 as one Ac/Nrecognin of the S. cerevisiae Ac/N-end rule pathway (39,59). We found that the degradation of the wild-type rat MLS rAANAT 3f (t1 ⁄2 Ͻ 5 min in wild-type S. cerevisiae) was not significantly impaired in doa10⌬ cells, in comparison with wild-type cells (Fig. 3, D and F). We did not examine, so far, the other yeast Ac/N-recognin, Not4 (see the Introduction) for its possible role in targeting the Nt-acetylated rat Ac-MLS rAANAT 3f in S. cerevisiae. We also do not know, thus far, whether Teb4, the mammalian counterpart of the yeast Doa10 Ac/N-recognin (60), is involved in the degradation of rat MLS rAANAT 3f in a homologous (mammalian) setting.
Degradation of Rat AANAT Is Proteasome-dependent and Does Not Require Lys-8 -The demonstrated degradation of two forms of the wild-type rat MLS rAANAT 3f (Ac-MLS rAANAT 3f and its non-Nt-acetylated counterpart) by the two branches of the N-end rule pathway ( Fig. 2) already implied the proteasome dependence of this degradation, given the known organization of the N-end rule pathway (Fig. 1, A and B). To verify this in S. cerevisiae by independent means, CHX-chases with rat MLS rAANAT 3f were carried out in wild-type versus pdr5⌬ cells in either the presence or absence of the MG132 proteasome inhibitor. (Cells lacking the transmembrane transporter PDR5 are more sensitive to MG132 (77, 78).) As expected, given the proteasome dependence of the N-end rule pathway (Fig. 1, A and B), the normally rapid degradation of the wild-type rat MLS rAANAT 3f was substantially inhibited in wild-type S. cerevisiae in the presence of 50 M MG132 (Fig.  4, A and B). This degradation was inhibited even more strongly in mutant pdr5⌬ cells under the same conditions, confirming the proteasome dependence of at least the bulk of MLS rAANAT 3f degradation (Fig. 4, A and B).
N-terminal regions of mammalian AANATs contain a highly conserved Lys residue (e.g. Lys-8 in rat AANAT and Lys-10 in human AANAT) that has been suggested as a potential ubiquitylation site and thus a determinant of the AANAT degron (33, 79 -81). To address this possibility, CHX-chases were performed in S. cerevisiae with wild-type MLS rAANAT 3ha and a mutant containing a Lys-8 3 Arg mutation ( MLS rAANAT 3ha K8R ). Wild-type MLS rAANAT 3ha and the mutant MLS rAANAT 3ha K8R were degraded at similar rates, indicating that the conserved lysine at this position (Lys-8 in rat MLS rAANAT) is not, by itself, a significant determinant of MLS rAANAT stability (Fig. 4C).
Degradation of Human AANAT in S. cerevisiae- Fig. 3 (A (lanes 6 -9) and B) shows the results of a CHX-chase, in wildtype yeast, with the 207-position residue (not counting the tag) (M)ST hAANAT 3f , the wild-type human AANAT C-terminally tagged with a triple-FLAG epitope. Parentheses around the superscript's N-terminal Met residue denote the fact that this Met is cotranslationally cleaved off by MetAPs. Whereas human (M)ST hAANAT 3f was relatively unstable in S. cerevisiae (t1 ⁄ 2 Ϸ 30 min), it was much longer-lived (including its higher zero-time, prechase level) than the wild-type rat MLS rAANAT 3f under the same conditions (Fig. 3, A and B).
As was also done with the wild-type rat MLS rAANAT 3f (Fig.  2C), we characterized the degradation of wild-type human (M)ST hAANAT 3f in S. cerevisiae strains deficient in specific components of the N-end rule pathway. The N-terminal Ser-Thr sequence of wild-type human (M)ST hAANAT 3f (after the cotranslational removal of the initially present N-terminal Met) made it likely that (M)ST hAANAT 3f was cotranslationally Ntacetylated in vivo (59,63). This modification of human (M)ST hAANAT 3f would make it a potential target for degradation by the Ac/N-end rule pathway (Fig. 1B). In contrast, the non-Nt-acetylated human (M)ST hAANAT 3f would not be expected to be recognized by either the Ac/N-end rule or the Arg/N-end rule pathway (Fig. 1A), in the latter case due to the absence of both a destabilizing N-terminal residue and a bulky hydrophobic (⌽) second residue. (A Met-⌽ N-terminal motif, containing a second-position ⌽ residue, can act as an N-degron recognized by the Arg/N-end rule pathway (60,82).
Given the much faster degradation of (M)ST hAANAT 3f in the absence of Naa10 (e.g. Fig. 3, C (lanes 1-4 versus lanes [5][6][7][8] and E) (i.e. in cells that would be incapable of Nt-acetylating (M)ST hAANAT 3f ), we also carried out CHX-chases, in wildtype S. cerevisiae, with (M)PT hAANAT 3f . In the latter mutant, the N-terminal Ser was replaced by the non-Nt-acetylatable Pro residue. In contrast to wild-type (M)ST hAANAT 3f in naa10⌬ cells, in which the (non-Nt-acetylated) (M)ST hAANAT became strikingly short-lived, the (non-Ntacetylatable) (M)PT hAANAT mutant was found to be longerlived, in wild-type cells, than the Nt-acetylatable wild-type (M)ST hAANAT 3f (Fig. 4, D and E). These results indicated that the observed accelerated degradation of the wild-type human  All quantified CHX-chase assays were carried out at least three times and yielded results within 10% of the data shown.
the Ub system that is distinct from the Arg/N-end rule pathway. (A proteolytic pathway in question would be distinct from the Arg/N-end rule pathway because (M)ST hAANAT was equally short-lived in single-mutant naa10⌬ and double-mutant naa10⌬ ubr1⌬ cells (Fig. 3, C-F).) In sum, the above Hsp90based mechanism (which remains to be addressed in the context of the (M)ST hAANAT protein) might underlie the strongly accelerated degradation of (M)ST hAANAT in naa10⌬ S. cerevisiae.
Degradation of Rat AANAT in Human HEK293T Cells-To compare the relative rates of degradation of N-terminal mutants of rat AANAT in human HEK293T cells, the wild-type rat MLS rAANAT 3f and its mutants (M)SMLS rAANAT 3f , (M)SI rAANAT 3f , and (M)PLS rAANAT 3f (see items i-iv at the beginning of "Results") were expressed in these cells using tran-sient transfection. Preliminary experiments with wild-type MLS rAANAT 3f encoded by a pcDNA3-based plasmid and expressed from the full-strength P CMV promoter indicated that the levels of expression of MLS rAANAT 3f were high enough to saturate or nearly saturate pathways that targeted MLS rAANAT 3f for degradation in HEK293T cells. Specifically, at those (high) levels of expression, wild-type rat MLS rAANAT 3f either was stable or was degraded slowly (data not shown).
We addressed this problem by constructing a set of truncated derivatives of the original P CMV promoter, aiming to attenuate its activity. The resulting nested set of truncated promoters, denoted as P CMVt1 , P CMVt2 , and P CMVt3 , comprised progressively shortened versions of the original P CMV . These truncations (Fig. 5A) were similar (although not identical) to a series of P CMV truncations described earlier by FIGURE 4. Degradation assays with a proteasome inhibitor and AANAT mutants in S. cerevisiae. A, lane 1, wild-type S. cerevisiae were transformed with vector (V) alone (control). Lanes 2-5, CHX-chase was performed at 30°C for the indicated times in wild-type S. cerevisiae with the wild-type rat MLS rAANAT 3ha . Lanes 6 -9, same as in lanes 2-5, but the CHX-chase was in the presence of the MG132 proteasome inhibitor (see "Experimental Procedures"). Lanes 10 -13, same as in lanes 2-5 but in pdr5⌬ S. cerevisiae lacking the efflux pump Pdr5. Lanes 14 -17, same as in lanes 10 -13 but in the presence of MG132. The bands of actin and MLS rAANAT 3ha are indicated on the left. B, quantification of data in A. C, CHX-chases were performed in wild-type S. cerevisiae with the wild-type rat MLS rAANAT 3ha (lanes 2-5) and with its Lys-8 3 Arg mutant MLS rAANAT 3ha K8R (see "Results"). The bands of tubulin and AANAT are indicated on the left. An asterisk on the right denotes a band of protein that cross-reacted with anti-HA antibody (the band is also present in lane 1, the vector-only control). D, lane 1, wild-type S. cerevisiae were transformed with vector (V) alone (control). Lanes 2-5, CHX-chase was performed at 30°C for the indicated times in wild-type S. cerevisiae with the wild-type human (M)ST hAANAT 3f . Lanes 6 -9, same as in lanes 2-5 but with the mutant human (M)PT hAANAT 3f . The bands of actin and AANAT are indicated on the left. E, quantification of data in A. All quantified CHX-chase assays were carried out at least three times and yielded results within 10% of the data shown. AUGUST 12, 2016 • VOLUME 291 • NUMBER 33 Promega Inc. (83). Each of the P CMVt1 , P CMVt2, and P CMVt3 promoter variants was cloned into pcDNA3, replacing fulllength P CMV .

Degradation of AANAT by the N-end Rule Pathway
The wild-type rat MLS rAANAT 3f was cloned into each of the above pcDNA3-based plasmids, and CHX-chases were performed in HEK293T cells transiently transfected with these constructs. Expression of MLS rAANAT 3f from the P CMVt1 and P CMVt2 promoters reduced time-zero levels of MLS rAANAT 3f (measured by immunoblotting at the beginning of CHX-chase) by ϳ70% and by ϳ97%, respectively, in comparison with the unmodified P CMV promoter (Fig. 5, A-C). With the most truncated promoter, P CMVt3 , no MLS rAANAT 3f could be detected at the same level of immunoblotting sensitivity (data not shown). The lower levels of MLS rAANAT 3f expression that have been attained through the use of the P CMVt1 and P CMVt2 promoters resulted in a strongly increased rate of the post-translational degradation of MLS rAANAT 3f (t1 ⁄ 2 Ͻ 2 h versus t1 ⁄ 2 Ͼ 8 h when MLS rAANAT 3f was expressed from the full-length P CMV promoter; Fig. 5, A-C). Treatment of HEK293T cells with the proteasome inhibitor MG132 during the chase increased the levels of MLS rAANAT 3f expressed from these promoters, thereby confirming that in each case, the bulk of degradation was proteasome-dependent (Fig. 5, A-C). The pcDNA3 CMVt1 plasmid, which contained P CMVt1 , one weakened version of the P CMV promoter, was chosen for experiments with HEK293T cells in this study.
We expected, a priori, that (M)SMLS rAANAT 3f would initiate from the first encoded Met residue, particularly because of the inclusion of the Kozak "optimal" upstream sequence in the relevant plasmids used for expression of test AANAT proteins. We could verify that expectation through chase-degradation assays using an affinity-purified antibody against the Nt-acetylated N terminus of MLS rAANAT 3f (Fig. 5D). We produced and purified this antibody as described under "Experimental Procedures." As shown in Fig. 5D, immunoblotting of SDS-PAGE-fractionated samples containing MLS rAANAT 3f or (M)SMLS rAANAT 3f with the anti-Ac-MLS rAANAT antibody generated a positive signal only in samples expressing MLS rAANAT 3f . If a construct expressing (M)SMLS rAANAT 3f had initiated translation (to a significant extent) from the second Met residue, the resulting translated protein would have been MLS rAANAT 3f . However, no signal was detected (using the above anti-Ac-MLS rAANAT antibody) with samples containing (M)SMLS rAANAT 3f (Fig. 5D), indicating negligible levels of translation initiation at the position 2 Met.
Although the mutant (M)PLS rAANAT 3f was partially stabilized relative to wild-type MLS rAANAT 3f , the former protein FIGURE 5. Weakened derivatives of the PCMV promoter in HEK293T cells and degradation assays using antibody to Nt-acetylated rat MLS rAANAT. A, a simplified diagram of the "wild-type" (unmodified) P CMV promoter in the cloning vector, with schematically illustrated binding sites for the NFB, SP1, and CREB transcriptional regulators. Two 5Ј-terminal truncations of wild-type P CMV , denoted as P CMVt1 and P CMVt2 , are also indicated. B, lane 1, HEK293T cells were transformed with vector (V) alone (control). CHX-chases were performed at 37°C for the indicated times in HEK293T cells with the wild-type rat MLS rAANAT 3f expressed either from the unmodified P CMV promoter (lanes 2-5), from the P CMVt1 promoter (lanes 7-10), or from the P CMVt2 promoter (lanes 12-15). Lanes 6, 10, and 16, same as lanes 5, 9, and 15 except that the MG132 proteasome inhibitor was present during 8-h CHX-chases in each case (see "Experimental Procedures" and "Results"). The bands of tubulin and MLS rAANAT 3f are indicated on the left. C, quantification of data in B. Symbols shown separately and framed in black squares correspond to altered (increased) levels of test proteins in the presence of MG132. All quantified CHX-chase assays were carried out at least three times and yielded results within 10% of the data shown. D, immunoblotting with affinity-purified antibody to Nt-acetylated MLS rAANAT (see "Experimental Procedures"). Lane 1, HEK293T cells were transformed with vector (V) alone (control) bearing the (weakened) P CMVt1 promoter. CHX-chases were performed at 30°C for the indicated times in HEK293T cells with the wild-type rat MLS rAANAT 3f (lanes 2-5) and its mutant (M)SMLS rAANAT 3f (lanes 7-10). Lanes 6 and 11, same as lanes 5 and 10 except that the MG132 proteasome inhibitor was present during 4-h CHX-chases in each case (see "Experimental Procedures"). The bands of rAANAT and tubulin (the latter a loading control) are indicated on the left.
was still significantly shorter-lived (t1 ⁄ 2 Ϸ 2 h) than the most stable (M)SI rAANAT 3f mutant (Fig. 6, A (lanes 17-21) and C). Although N-terminal Pro is not Nt-acetylated (59, 63) (this makes N-terminal Pro-bearing proteins invulnerable to the Ac/N-end rule pathway), the N-terminal Pro residue might still be (weakly) recognized by the Arg/N-end rule pathway in some sequence contexts. Specifically, our SPOT-type, peptide-based binding assays (84) suggested that the Ubr1 N-recognin of the Kzero (lanes 7-10). Lanes 6 and 11, same as lanes 5 and 10 except that MG132 was present during a 4-h CHX-chase in each case. Right, same as in the left panel, but both the wild-type MLS rAANAT 3f and its lysine-lacking counterpart were C-terminally tagged with a triple-FLAG tag, which contains a Lys residue, in contrast to the HA tag, which lacks lysines (see "Results" and "Discussion"). C, quantification of data in A. D, quantification of data in B. Symbols shown separately and framed in black squares or rectangles correspond to altered levels of test proteins in the presence of MG132. All quantified CHX-chase assays were carried out at least three times and yielded results within 10% of the data shown. E, immunoblot analyses (IB), using anti-Ub antibody, of total, input (before immunoprecipitation) extracts from HEK293T cells that had been transfected either with vector alone (lanes 1 and 2) or with plasmid expressing the wild-type rat MLS rAANAT 3ha (lanes 3 and 4) or with plasmid expressing its lysine-lacking counterpart MLS rAANAT 3ha Kzero (lanes 5 and 6). The MG132 proteasome inhibitor was either omitted (lanes 1, 3, and 5) or added to HEK293T cells (to a final concentration of 10 M) 6 h before preparation of extracts (lanes 2, 4, and 6). The bottom panel shows the results of detecting non-ubiquitylated MLS rAANAT 3ha (absent in the control lanes 1 and 2) with anti-HA antibody. F, same as in E but immunoblot analyses (using anti-Ub antibody) of the input samples after their immunoprecipitation (IP) with anti-HA antibody (followed by SDS-PAGE) to isolate and detect ubiquitylated AANAT 3ha species.
S. cerevisiae Arg/N-end rule pathway (and, by inference, mammalian Ubr1 and Ubr2 as well) may weakly bind to N-terminal Pro when this residue is followed by a hydrophobic residue. 4 Thus, one interpretation of the "residual" instability of the rat (M)PLS rAANAT 3f mutant (in comparison with the longest-lived (M)SI rAANAT 3f mutant) is that the (non-Nt-acetylated) N-terminal sequence Pro-Leu of (M)PLS rAANAT 3f is weakly targeted by the Arg/N-end rule pathway, in contrast to the absence of targeting of the N-terminal Ser-Ile sequence of the (M)SI rAANAT 3f mutant (Fig. 6, A and C; also see Fig. 2, B and D).
As mentioned above, we found that transient transfectionbased expression of AANAT from the full-strength P CMV promoter could be high enough to result in saturation or near saturation of the relevant proteolytic pathways. Although the use of the considerably weaker P CMVt1 promoter at least partially solved this problem (see above), additional studies would be required to make sure that pathway saturation effects were consistently negligible with longer-lived AANAT derivatives that were expressed from P CMVt1 . That said, the preponderance of our data makes it nearly certain that the observed differences in time-zero (before chase) levels of AANAT proteins were caused largely by differences in the rates of their degradation.

The Bulk of Degradation of Rat AANAT Is Mediated by Polyubiquitylation of Its Lys Residue(s)-An earlier study, in which
the wild-type rat MLS rAANAT was apparently not stabilized by conversion of its Lys residues to Arg, was based on measurements of steady-state levels of MLS rAANAT in HEK293 cells that had been treated or left untreated with the proteasome inhibitor MG132 (37). To address this issue more directly, we generated a triple-HA-tagged mutant, denoted MLS rAANAT 3ha Kzero , of the wild-type rat MLS rAANAT in which all four of its Lys residues were converted to Arg. (A FLAGbased epitope tag could not be employed in these experiments, because FLAG contains a Lys residue that could potentially serve as a target for polyubiquitylation.) CHX-chases in HEK293T cells (performed as described above) with the wild-type MLS rAANAT 3ha versus its lysine-lacking MLS rAANAT 3ha Kzero mutant indicated that MLS rAANAT 3ha Kzero was nearly completely stable during the chase (t1 ⁄ 2 Ͼ Ͼ 4 h), whereas the wild-type MLS rAANAT 3ha was degraded, with t1 ⁄ 2 Ϸ 2 h (Fig. 6, B  and D). A treatment with MG132 during CHX-chase strongly stabilized the wild-type MLS rAANAT 3ha , confirming that its degradation is proteasome-dependent (Fig. 6, B (left, lanes 2 and 6) and D). In contrast, the MG132 treatment during CHX-chase of the lysine-lacking MLS rAANAT 3ha Kzero mutant produced at most a marginal stabilizing effect, suggesting that even if there is a small (residual) amount of lysine-independent degradation of MLS rAANAT 3ha Kzero , such a degradation would still require the proteasome (Fig. 6, B (left, lanes 7 and 11) and D).
In a different approach to the same problem, HEK293T cells were transiently transfected with the pcDNA3 vector or with plasmids expressing either MLS rAANAT 3ha or MLS rAANAT 3ha Kzero . Cells were then treated for 6 h with either MG132 or an equivalent volume of DMSO (in which the stock solution of MG132 was made). Thereafter, cell extracts were immunoprecipitated with anti-HA magnetic beads, and the immunoprecipitates were fractionated by SDS-PAGE, followed by immunoblotting with anti-HA and anti-ubiquitin antibodies. Polyubiquitin chains (apparently linked to MLS rAANAT 3ha ) could be readily detected in fractionated immunoprecipitates of extracts from cells that expressed MLS rAANAT 3ha (but, crucially, not from cells that expressed an empty vector), and the levels of polyubiquitin chains were significantly increased in the presence of MG132 (Fig. 6F,  lanes 1-4). In contrast, significantly lower amounts of polyubiquitin chains were observed in fractionated immunoprecipitates of extracts from cells that expressed the lysine-lacking MLS rAANAT 3ha Kzero (Fig. 6F, lane 5). Moreover, the presence of MG132 did not increase the low levels of detected polyubiquitin chains (Fig. 6F, lane 5 versus lane 6). The latter result suggested that (residual) polyubiquitin chains observed in cells that expressed MLS rAANAT 3ha Kzero may be of a kind (e.g. see Ref. 85) that do not contribute to the proteasome-dependent degradation of MLS rAANAT 3ha Kzero . In agreement with this interpretation, the lysine-lacking MLS rAANAT 3ha Kzero mutant was expressed at significantly higher steady-state levels than the wild-type rat MLS rAANAT 3ha in the absence of MG132 (in DMSO-treated cells) (Fig. 6E, bottom), suggesting that the absence of Lys residues stabilized MLS rAANAT 3ha Kzero against degradation throughout life histories of MLS rAANAT 3ha Kzero molecules, including, possibly, their degradation during or immediately after their synthesis.
Degradation of Human AANAT in Human HEK293T Cells-Degradation of the wild-type human (M)ST hAANAT 3f and its N-terminal Pro residue-bearing mutant (M)PT hAANAT 3f (see items vi and vii at the beginning of "Results") were examined by CHX-chases in HEK293T cells. In agreement with the findings in S. cerevisiae (Figs. 2 and 3), the wild-type human (M)ST hAANAT was longer-lived than the wild-type rat MLS rAANAT 3f in HEK293T cells (compare Fig. 7A with Fig.  6A). Interestingly, the mutant human (M)PT hAANAT 3f was less stable than the wild-type human MLS rAANAT 3f during the course of a 2-h chase in HEK293T cells (Fig. 7, A and C).
We also asked whether the (relatively slow) degradation of the wild-type human (M)ST hAANAT 3f (Fig. 7, A and C) required both its lysine-dependent ubiquitylation and the proteasome. CHX-chases were carried out with (M)ST hAANAT 3ha and its lysine-lacking mutant (M)ST hAANAT 3ha Kzero . The latter protein was stable during the chase (t1 ⁄ 2 Ͼ Ͼ 4 h), whereas wild-type (M)ST hAANAT 3ha exhibited a weak but observable instability (Fig. 7, B and D). Treatment with the MG132 proteasome inhibitor increased the level of the wild-type human (M)ST hAANAT 3ha during CHX-chase, confirming proteasome dependence of its (slow) degradation (Fig. 7, B (lanes 3 and 7) and D).

Discussion
The ϳ23-kDa AANAT (also called serotonin N-acetyltransferase) is apparently universal among animals and is present in plants as well. AANAT converts the neurotransmitter serotonin to NAS, a regulatory compound in its own right and the immediate precursor of melatonin, a circulating hormone that regulates sleep and other circadian processes in vertebrates. The levels of melatonin and NAS are modulated by oscillatory circadian circuits and also impact those circuits, in addition to other effects of these compounds (see the Introduction). Hence the importance of temporal control of the AANAT enzyme. The expression of AANAT is regulated through both transcriptional and post-translational mechanisms. Details of these mechanisms differ among vertebrates. In addition, although AANAT enzymes from different species are highly sequelogous (71) throughout their ORFs, the sequences of the first ϳ10 residues of AANAT tend to differ even among relatively closely related mammals, let alone other animals (Fig. 1C). This pattern of variation suggests a blend of adaptive (function-linked, natural selection-based) changes in N-terminal sequences of AANATs versus an unknown (extent-wise) but apparently strong flux of quasi-neutral, drift-mediated changes of these N-terminal sequences on evolutionary time scales.
In the present study, we approached the proteolysis-based regulation of rat and human AANATs (Fig. 1C) by expressing them and their mutants not only in mammalian cells but also in the yeast S. cerevisiae. This strategy made possible the use of yeast genetics to illuminate specific proteolytic pathways involved, in parallel with experiments in mammalian cells.
We found that the wild-type rat AANAT ( MLS rAANAT) is targeted for degradation by two complementary branches of the N-end rule pathway. Specifically, we showed that the Ntacetylated Ac-MLS rAANAT is destroyed through the recognition of its Nt-acetylated N-terminal Met residue by the Ac/Nend rule pathway, whereas the non-Nt-acetylated MLS rAANAT is targeted for degradation by the Arg/N-end rule pathway, which recognizes the unacetylated N-terminal Met-Leu sequence of MLS rAANAT.
The N-terminal sequence of the wild-type human (M)ST hAANAT ((Met)-Ser-Thr) is different from the Met-Leu-Ser sequence of rat MLS rAANAT (Fig. 1C). We found that the wild-type human (M)ST hAANAT is considerably longer-lived than its rat MLS rAANAT counterpart and does not appear to be an efficacious N-end rule substrate. Together, these and related results (Figs. 2-7) indicated both a major involvement of the N-end rule pathway in the control of rodent (specifically rat) AANAT and substantial differences in the regulation of rodent and human AANATs that stem from differences in their N-terminal sequences.
The observed co-targeting of the two "complementary" forms of the rat MLS rAANAT by the two "complementary" branches of the N-end rule pathway (Figs. 2 and 3) is in agreement with the earlier finding that the N-terminal region of MLS rAANAT, and specifically its Leu-2 residue, are an important determinant of the observed metabolic instability of this AANAT (37). The cited study also suggested that either the Lys residues of MLS rAANAT or its surface-exposed Cys residues were not essential for its proteasome-mediated degradation (37). In the present work, we compared the degradation of wildtype MLS rAANAT and its lysine-lacking MLS rAANAT 3ha Kzero mutant using CHX-chases. Our results showed that the C-terminally triple-HA-tagged MLS rAANAT 3ha Kzero mutant was completely stabilized in human HEK293T cells, in contrast to instability of wild-type MLS rAANAT, indicating that at least some Lys residues of MLS rAANAT were required for its degradation (Fig. 6, B and D).
A possible explanation of the above discrepancy stems from different choices of C-terminal epitope tags in the earlier and the present study. Specifically, Huang et al. (37) employed MLS rAANAT C-terminally tagged with c-Myc epitope (EQKLI-SEEDL), which contains a Lys residue. In the present work, we used a triple-HA tag, which lacks Lys residues (the sequence of  6 -9). B, lane 1, fluorescently labeled molecular mass markers, with their masses, in kDa, indicated on the left. Lane 2, HEK293T cells were transformed with vector (V) alone (control). CHX-chases were performed at 37°C for the indicated times in HEK293T cells with the wild-type human (M)ST hAANAT 3ha (lanes 3-6) and with its lysine-lacking mutant (M)ST hAANAT 3ha Kzero (lanes 8 -11). Lanes 7 and 12, same as lanes 6 and 11, except that the MG132 proteasome inhibitor was present during a 4-h CHX-chase in each case (see "Experimental Procedures"). The bands of actin and AANAT are indicated on the right. Asterisks on the right in A and B denote a protein band (possibly the same protein and possibly a derivative of AANAT) that reacted with both anti-FLAG antibody in A and anti-HA antibody in B. C, quantification of data in A. D, quantification of data in B. E, quantification of data in Fig. 6B (right), similar to the analogous quantifications described in the legend to Fig.  6, C and D. All quantified CHX-chase assays were carried out at least three times and yielded results within 10% of the data shown. AUGUST 12, 2016 • VOLUME 291 • NUMBER 33

JOURNAL OF BIOLOGICAL CHEMISTRY 17189
single HA is YPYDVPDYA). Previous analyses of N-degrons have shown that the targeting apparatus of the Arg/N-end rule pathway selects a substrate's internal lysine as a polyubiquitylation site through a process in which different (spatially competing, conformationally mobile) Lys residues can be chosen stochastically, depending on their ability to become transiently close to the substrate-bound E2-E3 complex (41,51,86). Thus, the presence of a Lys residue in the C-terminal c-Myc epitope of the earlier study (37) might have rendered the otherwise lysinelacking MLS rAANAT mutant still targetable by the N-end rule pathway. Indeed, tellingly, our initial experiments with lysinelacking mutants of MLS rAANAT have employed, inadvertently, a lysine-containing C-terminal FLAG tag instead of the lysinelacking HA tag. Remarkably, the otherwise lysine-lacking but FLAG-tagged MLS rAANAT 3f Kzero mutant was found to be unstable (Figs. 6B (right) and 7E), in contrast to the actually (completely) lysine-lacking, HA-tagged MLS rAANAT 3ha Kzero mutant (Fig. 6, B (left) and D), in agreement with the above tag-based explanation.
In addition to polyubiquitylation and degradation, AANAT is also regulated through a site-specific phosphorylation, a modification that stimulates the binding of AANAT to at least one of the 14-3-3 chaperone-like proteins (they are encoded by a family of several 14-3-3 genes in mammals). This interaction with 14-3-3 both protects (phosphorylated) AANAT from degradation and enhances its catalytic activity. Rat, human, and other AANATs contain two previously mapped phosphorylation sites, Thr-29 and Ser-203 (position numbers refer to the sequence of rat MLS rAANAT). Phosphorylation at Thr-29 is conserved in all examined mammalian AANATs (31). This phosphorylation, by PKA and/or by PKC, augments the interaction between AANAT and 14-3-3 both in vitro and in vivo (1,5,37). By expressing AANATs not only in mammalian cells but also in the heterologous setting of S. cerevisiae, we could analyze the degradation of AANATs in the possible absence of phosphorylation at Thr-29 (or its equivalent) and also in the possible absence of AANAT interactions with 14-3-3 proteins. Whereas S. cerevisiae has two 14-3-3 proteins (87), there is no evidence, so far, that bears on their binding, in vivo, to a mammalian AANAT.
Our attempts to observe a phosphorylation-induced in vivo binding of the wild-type rat MLS rAANAT either to an endogenous yeast 14-3-3 protein or to the (overexpressed) human 14-3-3 protein were unsuccessful so far, despite a variety of tried approaches. 5 Nevertheless, the likely conceptual benefit of achieving a phosphorylation-inducible interaction between a mammalian AANAT and a specific 14-3-3 protein, in a setting of greatly reduced complexity of 14-3-3s (in comparison with the large set of their mammalian counterparts), justifies further work in this direction. 14-3-3 proteins are abundant, broadly expressed, multifunctional chaperone-like proteins that modulate catalytic activities of enzymes to which 14-3-3s bind and also regulate protein-protein interactions and subcellular targeting of specific proteins (88 -90). It is possible, indeed likely, that 14-3-3 proteins may bind, in a regulated manner, not only to, for example, rat MLS rAANAT but to other N-end rule substrates as well, thereby controlling the rates of degradation of such substrates in ways that remain to be explored.
Recently, the naturally occurring mutant human protein RGS2 Q2L was shown to be a conditionally short-lived substrate of both the Ac/N-end rule pathway and the Arg/N-end rule pathway (60). This clinically relevant mutant of RGS2 (a negative regulator of specific G proteins) was identified in a cohort of hypertensive Japanese patients (91) (also see the Introduction). In the present study, we demonstrated that the otherwise unrelated rat MLS rAANAT protein, whose N-terminal Met-Leu sequence is identical to that of RGS2 Q2L , is also targeted for degradation by both branches of the N-end rule pathway (Figs.  1 and 2). These results further expanded the set of identified mammalian N-end rule substrates of the Met-⌽ type.
Interestingly, RGS2, an N-end rule substrate (60), has been identified as a physiologically relevant inhibitor of the AANATdependent melatonin production (92). Light-stimulated, norepinephrine-induced increases in the levels and activity of AANAT are mediated by G-protein-coupled adrenergic receptors, whose activation increases intracellular cAMP, and subsequently the levels of AANAT as well, at least in part through an increased phosphorylation of AANAT and the resulting induction of its binding to protective 14-3-3 proteins (5,28). While acting to increase the levels and activity of AANAT (which is required for the synthesis of NAS and melatonin), cAMP also acts as a negative feedback inhibitor of melatonin production through up-regulation of Rgs2 transcription. The resulting increase in RGS2 inhibits the activity of G proteins that are coupled to the norepinephrine-responsive receptor, a negative feedback circuit that acts to decrease cAMP levels (92). AANAT results of the present study, together with the earlier evidence that RGS2 is both a conditionally short-lived N-end rule substrate and an (indirect) down-regulator of AANAT (60,92), indicate that the N-end rule pathway is a pleiotropic controller of circuits that either directly or indirectly modulate AANAT.
Previous studies of AANAT demonstrated that its transcriptional regulation can be different in different species, but it has been conjectured that mechanisms and regulation of AANAT degradation would be more conserved in evolution (11, 34 -36). As shown in the present study, the rates of degradation of the human and rat AANAT proteins are quite different. Moreover, whereas rat MLS rAANAT is destroyed by the N-end rule pathway (Figs. 2 and 6), human (M)ST hAANAT is substantially resistant to this pathway (Figs. 3 and 7). Nevertheless, if the identification of RGS2 as a feedback inhibitor of AANAT expression in the rat pineal gland (92) could also be shown to extend to human RGS2, the N-end rule pathway would be a regulator of AANAT in humans as well, albeit an indirect one.
An earlier study of human (M)ST hAANAT described its apparent degradation by the proteasome but did not detect its ubiquitylation (33). In that study, the instability of human (M)ST hAANAT was inferred from increases in its steady-state levels in the presence of a proteasome inhibitor. In the present work, we employed, in particular, CHX-chases to show that although human (M)ST hAANAT is degraded by the proteasome, it is substantially longer-lived than rat MLS rAANAT 5 B. Wadas and A. Varshavsky, unpublished data.

Degradation of AANAT by the N-end Rule Pathway
under the same conditions, either in mammalian cells or in S. cerevisiae (Figs. 2-7).
In sum, our results indicate both a major involvement of the N-end rule pathway in the control of rodent AANATs and substantial differences in the regulation of rodent and human AANATs that stem from differences in their N-terminal regions (Fig. 1C). What is the cause of the remarkable variability of N-terminal sequences of AANATs during mammalian (and, more generally, animal) evolution? Given the extent of variability (Fig. 1C), it is nearly certain that a substantial fraction of these changes resulted from a quasi-neutral, unselected genetic drift (93). At the same time, the broad and biologically relevant variation in detailed circadian rhythms among different animal species (due to specific ecological and physiological adaptations) and the role of AANAT in these rhythms suggest that the seeming randomness of the N-terminal sequences of AANATs (Fig. 1C) may be obscuring, so far, a set of adaptive (selected) changes in these sequences over evolutionary time scales.

Experimental Procedures
Yeast Strains, Media, and Genetic Techniques-Standard yeast genetic techniques were used (94 -96). S. cerevisiae strain BWY29 was constructed by transforming the strain CHY345 (ubr1⌬::LEU2 in the strain background of BY4742) with a PCRamplified DNA fragment that encoded the selection marker KanMX6 (making cells resistant to kanamycin) and was targeted to the 5Ј-and 3Ј-flanking regions of the NAA10 gene, thereby replacing the ORF of NAA10 with KanMX6 (Table 1). S. cerevisiae strain JOY487 was made by transforming BY4742 with a PCR-amplified DNA fragment that encoded the natNT2 marker (making cells resistant to nourseothricin) and was targeted to the 5Ј-and 3Ј-flanking regions of PDR5, thereby replacing the ORF of PDR5 with natNT2 ( Table 1). The resulting pdr5⌬ cells were used in experiments that involved the MG132 proteasome inhibitor (Fig. 4, A and B). Other S. cerevisiae strains used in this study were constructed previously and are cited in Table 1. S. cerevisiae were transformed using the LiAc/PEG method (97). S. cerevisiae media included YPD medium, synthetic complete medium, and synthetic drop-out medium (95,96).
Plasmids, cDNAs, and Primers-Turbo Escherichia coli (New England Biolabs) was used for cloning and maintaining plasmids. Phusion High-Fidelity DNA polymerase (New England Biolabs) was used for carrying out PCR. Nucleotide sequences of all plasmids were verified by DNA sequencing. The plasmids and PCR primers used in this study are described in Tables 2  and 3, respectively.
The low copy pBW395 and pBW419 plasmids expressed the wild-type human (M)ST hAANAT C-terminally tagged with a triple-FLAG epitope ( (M)ST hAANAT 3f ) and its N-terminal mutant (M)PT hAANAT 3f , respectively, from the P GAL1 promoter. To construct these plasmids, the human AANAT ORF from Genscript clone OHu55705 (Genscript, Piscataway, NJ) was amplified using primers BW705, BW711, BW712, and BW244 or primers BW771, BW711, BW712, and BW244, respectively ( Table 3). The PCR-amplified DNA fragments were digested with EcoRI and HindIII and were ligated into EcoRI/HindIII-cut pRS416 GAL1 (99). All final DNA constructs were verified by DNA sequencing. See items i-x at the beginning of "Results" for information about rat and human AANAT notations.
For expression of the above AANAT-encoding ORFs in human HEK293T cells, these ORFs were cloned into a modified expression vector derived from pcDNA3. That (modified) vector, termed pcDNA3 CMVt1 , contained a truncated (weakened) version of the P CMV promoter, termed P CMVt1 . The pcDNA3 CMVt1 plasmid was constructed by amplifying, using PCR, the 276-bp fragment that encompassed 104 bp at the 3Ј-end of the P CMV promoter region in pcDNA3, its T7 promoter site, and its multiple cloning site, using primers BW375 and BW376 (Table 3). The amplified DNA fragment was digested with NruI and XbaI and ligated into NruI/XbaI-cut pcDNA3, yielding the plasmid pBW173 (Table 2). Its truncated P CMVt1 promoter retained several proximal promoter elements, including the consensus NFB site, the CREB-binding site, and two SP1/SP3-binding sites, but lacked more distal enhancer elements. This resulted in a ϳ70% lower activity of the P CMVt1 promoter in HEK293T cells, in comparison with P CMV (Fig. 5, A-C).
The pcDNA3 CMVt1 -based plasmids pBW206, pBW211, pBW212, and pBW213 expressed, in HEK293T cells, either the wild-type rat MLS rAANAT 3f or its N-terminal mutants (M)SI rAANAT 3f , (M)SMLS rAANAT 3f , and (M)PLS rAANAT 3f from the P CMVt1 promoter. To construct these plasmids, Aanat ORFs were amplified from the plasmids pBW105, pBW106, This study pBW107, and pBW209 using primers BW352, BW353, BW354, BW413, and BW355, respectively ( Table 3). The amplified DNA fragments were digested with EcoRI and XhoI and ligated in EcoRI/XhoI-cut pBW173 ( Table 2). All pcDNA3-based constructs designed for expression in HEK293T cells contained the Kozak sequence 5Ј-ACC immediately upstream of the start codon. pBW477, which expressed the MLS rAANAT 3ha from the P CMVt1 promoter, was constructed by using pBW135 as a PCR template and the primers BW352/BW340. The resulting PCRamplified DNA fragment was digested with EcoRI and XhoI and ligated into EcoRI/XhoI-cut pBW173 ( Table 2). pBW478 expressed MLS rAANAT 3ha Kzero (i.e. the lysine-lacking derivative of MLS rAANAT 3ha ). It was constructed by using pBW135 as a PCR template to generate four separate DNA fragments containing lysine-to-arginine mutations. The primer pairs were 704/784, 785/786, 787/788, and 789/340 (Table 3). These fragments were then assembled into a contiguous ORF using PCR. The resulting DNA fragment, encoding lysine-to-arginine mutations at each of the four lysines contained in the wild-type rat MLS rAANAT 3f , was digested with EcoRI and XhoI, and ligated into EcoRI/XhoI-cut pBW173.
pBW466 and pBW467, the pcDNA3-based counterparts of pBW395 and pBW419 (Table 2) were generated by PCR using pBW395 and pBW419 as templates and the primers BW866/ BW355 and BW868/BW355, respectively (Table 3). PCR-amplified DNA fragments were digested with EcoRI and XhoI and ligated into EcoRI/XhoI-cut pBW173 (Table 2). Cell Culture-The HEK293T cell line (derived from human embryonic kidney cells) was obtained from American Type Culture Collection (ATCC, Manassas, VA) and was grown at 37°C in 5% CO 2 in DMEM supplemented with 10% FBS (Gemini Bio-Products, West Sacramento, CA) and penicillin/streptomycin (100 units/ml; Hyclone). Cells were transfected using Lipofectamine-2000 (Invitrogen) according to the manufacturer's protocol.
Antibody Specific for the Nt-acetylated Form of Rat AANAT ( MLS rAANAT)-Unpurified rabbit polyclonal antisera against the synthetic peptide Ac-MLSIHPLKPEAC were produced by Abgent (San Diego, CA) essentially as described for other peptide-mediated polyclonal antibodies produced in studies by the Varshavsky laboratory (e.g. see Refs. 59 and 66). The sequence of the Nt-acetylated peptide immunogen comprised the first 11 residues of MLS rAANAT ( Fig. 1C; see below), followed by the C-terminal Cys residue that was used to conjugate the peptide to keyhole limpet hemocyanin carrier protein. The antiserum was affinity-purified in two sequential steps. First, the antibody sample was incubated with the C-terminally immobilized original immunogen Ac-MLSIHPLKPEAC. The peptide-bound fraction was eluted as described previously (59,66). That fraction was thereafter "negatively" purified against the otherwise identical but non-Nt-acetylated immobilized MLSIHPLKPEAC peptide (the peptides were immobilized using a sulfhydryl-coupling resin (Abgent)), collecting, this time, the unbound fraction. The resulting antibody sample was used to detect, selectively, the Nt-acetylated form of rat AANAT (see "Results" and Fig.  5D). The fraction of once purified antibody that was bound to the non-Nt-acetylated MLSIHPLKPEAC peptide was also eluted and retained and was used as an antibody that could recognize the N-terminal region of rat AANAT irrespective of its Nt-acetylation state.
Cycloheximide Chase Assay in S. cerevisiae-CHX-chase assays in S. cerevisiae were performed largely as described (39,59). Briefly, S. cerevisiae cells were grown in overnight cultures, and expression of epitope-tagged test proteins was induced by the addition of 2% galactose (for plasmids containing the P GAL1 promoter) or 0.1 mM CuSO 4 (for plasmids containing the P CUP1 promoter (100)). 3 h post-induction, CHX was added to a final concentration of 0.1 mg/ml. Samples were collected at the indicated time points, centrifuged at 11,200 ϫ g for 2 min to pellet the cells, and snap-frozen in liquid nitrogen. Proteins were extracted by resuspending the pellets in 1 ml of 0.2 M NaOH and incubating on ice for 20 min (101). Cells were then pelleted and resuspended in 50 l of SUMEB loading buffer (1% SDS, 8 M urea, 10 mM EDTA, 0.01% bromphenol blue, 10 mM MOPS, pH 6.8), followed by heating at 95°C for 10 min. Samples were then centrifuged at 11,200 ϫ g for 5 min, followed by SDS-PAGE (with 10 l loaded per well) using 4 -12% NuPAGE gels (Invitrogen). Fractionated proteins were transferred onto nitrocellulose membranes for immunoblotting analyses (see below).
Cycloheximide Chase Assay in Mammalian Cell Lines-HEK293T cells were transfected with 4 g of plasmid DNA per well (10-cm 2 surface area) in 6-well plates using Lipofectamine-2000 (Invitrogen) according to the manufacturer's protocol. 24 h after transfection, cells were treated with CHX (0.1 mg/ml) to initiate a chase. All cells from one well were collected at each indicated time point by washing cells quickly in ice-cold PBS and then rapidly scraping cells into 1.5-ml tubes, pelleting them by centrifugation at 4°C (at 10,000 ϫ g for 1 min), and snapfreezing the pellets in liquid nitrogen. Cells in these samples were later lysed by the addition of 0.2 ml of "mammalian lysis buffer" (1% Nonidet P-40, 0.15 M NaCl, 50 mM Tris-HCl, pH 7.5) containing 1ϫ Roche Complete Protease Inhibitor Mixture and brief sonication, followed by centrifugation at 11,200 ϫ g for 10 min. Total protein concentration in the supernatants was measured by the bicinchoninic acid (BCA) assay (Thermo Fisher Scientific). 30 g of total protein in a thus prepared extract in lithium dodecyl sulfate (LDS)-sample buffer (in a volume of 45 l) were heated at 70°C for 10 min, followed by LDS-PAGE on a 4 -12% BisTris NuPAGE gel (Invitrogen) and subsequent transfer onto nitrocellulose membranes for immunoblotting.
Immunoprecipitation of Polyubiquitylated AANAT-HEK293T cells were transfected with 4 g of plasmid DNA per well (10-cm 2 surface area) in 6-well plates using Lipofectamine-2000 (Invitrogen) according to the manufacturer's protocol. 24 h after transfection, cells were treated with the proteasome inhibitor MG132 (final concentration of 10 M; AG Scientific, San Diego, CA) or with an equivalent volume of DMSO in which the 10 mM stock solution of MG132 was made, as indicated. 6 h post-treatment, cells were collected by washing them quickly on a plate in ice-cold PBS and then rapidly scraping cells into 1.5-ml tubes and pelleting them by centrifugation at 4°C at 10,000 ϫ g for 1 min. Pellets were resuspended in 0.1 ml of buffer containing 1% SDS and 50 mM Tris (pH 7.5) and heated at 95°C for 10 min. After boiling, samples were diluted with 1 ml of buffer TNN (0.5% Nonidet P-40, 0.25 M NaCl, 5 mM EDTA, 50 mM Tris, pH 7.5) containing 1ϫ Roche Complete Protease Inhibitor Mixture, 20 mM N-ethylmaleimide (Sigma-Aldrich), and 50 M PR-619 (LifeSensors, Malvern, PA) as inhibitors of deubiquitylation. 25 l of anti-HA magnetic beads were then added to each sample, and samples were incubated at 4°C overnight, followed by three washes in TNN, 1 wash in 10 mM Tris-HCl, pH 8.5, and elution with 50 l of 0.1 M glycine, pH 2.0, by incubation at room temperature for 10 min. After elution, the samples were neutralized and then heated at 70°C in LDS-PAGE sample buffer, followed by LDS-PAGE on a 4 -12% NuPAGE gel. Fractionated proteins were then electroblotted onto nitrocellulose membranes. The resulting membranes (containing transferred proteins) were autoclaved as described previously (102) before "downstream" immunoblotting procedures with either anti-HA or anti-ubiquitin antibodies to increase the sensitivity of detection of polyubiquitin chains.